{
  "id": 43539,
  "title": "Supportive Data Preprocess within Low Memory Configuration",
  "url": "/competitions/favorita-grocery-sales-forecasting/discussion/43539",
  "author_name": "Robin Hoo",
  "post_date": "2017-11-16T03:22:11.945000",
  "votes": 0,
  "comment_count": 0,
  "views": 0,
  "content": "<p>As the dataset size is large and when added in the date type will increase the size much bigger, here I found a way to convert the date into integer and with relevant information not lost. Meanwhile, some memory reduce method also used to keep the full data size(train+test) smaller enough to run in 8G Mac OS.</p>\n\n<p>You may refer to the code in kernel <a href=\"https://www.kaggle.com/robinhoo1973/supportive-data-preprocess-with-low-memory\">https://www.kaggle.com/robinhoo1973/supportive-data-preprocess-with-low-memory</a></p>",
  "messages": [
    {
      "id": 244344,
      "postDate": "2017-11-16T03:22:11.947Z",
      "content": "<p>As the dataset size is large and when added in the date type will increase the size much bigger, here I found a way to convert the date into integer and with relevant information not lost. Meanwhile, some memory reduce method also used to keep the full data size(train+test) smaller enough to run in 8G Mac OS.</p>\n\n<p>You may refer to the code in kernel <a href=\"https://www.kaggle.com/robinhoo1973/supportive-data-preprocess-with-low-memory\">https://www.kaggle.com/robinhoo1973/supportive-data-preprocess-with-low-memory</a></p>",
      "rawMarkdown": "As the dataset size is large and when added in the date type will increase the size much bigger, here I found a way to convert the date into integer and with relevant information not lost. Meanwhile, some memory reduce method also used to keep the full data size(train+test) smaller enough to run in 8G Mac OS.\n\nYou may refer to the code in kernel https://www.kaggle.com/robinhoo1973/supportive-data-preprocess-with-low-memory\n"
    }
  ],
  "comments": [],
  "raw_markdown_by_id": {
    "244344": "As the dataset size is large and when added in the date type will increase the size much bigger, here I found a way to convert the date into integer and with relevant information not lost. Meanwhile, some memory reduce method also used to keep the full data size(train+test) smaller enough to run in 8G Mac OS.\n\nYou may refer to the code in kernel https://www.kaggle.com/robinhoo1973/supportive-data-preprocess-with-low-memory\n"
  }
}